Tensile Strength Testing

[rsterne]

I recently purchased a Digital Force Gauge from Amazon, so that I could do some testing to better understand the tensile strength of various filament materials, and the effect of changes in printer settings on that strength....



I downloaded some test samples from Thingiverse, both for printing flat to test the X-Y strength, and for printing vertical to test the Z (layer bonding) strength.... The first ones I printed were the flat ones, and I did 3 versions, with 2, 4 and 8 walls, all at 100% fill, to see if having more filament extruded parallel to the applied force was better.... The Infill is always at 100%....



You can see the difference in the above photo.... However, when I measured the parts, the square test section, which was supposed to be 0.200" square was a bit oversize (0.206-0.210"), with an area averaging 0.0436 sq.in. (9% oversize) over 6 samples, and the holes were about 0.360" diameter.... I wanted to use 3/8" bolts/pins to pull them apart, and if the area of the test section is (0.2 x 0.2) = 0.04 sq.in. all I need to do is set the force gauge to read in lbs. and multiply by (1 / 0.04) = 25 to get the tensile strength in psi.... So, I used TinkerCad to redesign them, and design a matching vertically printed test bar, so that I can fine tune the dimensions the way I want.... Here is the first set just printed....



The vertically printed one is 10mm thick instead of 5mm at the ends (as was the original) to make sure the holes don't tear out, but the test section is still 5mm square (as designed on TinkerCad).... The holes were designed at 10mm, on 33mm centers, and the ends of the bars are 22mm wide, and the bars are 55mm long.... The test sections (which are 6mm long) are much closer to what I want, the horizontal one measures 0.194 x 0.204" for an area of 0.396 sq.in, and the vertical one measures 0.197 x 0.201" which is also 0.396 sq.in. (within 1% of desired)... The holes are the correct size, only needing a quick deburr on the corners to fit perfectly on the shank of a 3/8" bolt....



If you haven't got a deburring tool, they are cheap, and if you get the plastic blades are great for deburring holes or corners on 3D prints.... The blade spins in the handle, and it comes with 10 spare blades that store inside the handle.... 



The test sections are now close enough to the correct size (0.04 sq.in) that they can be fine tuned using the slicer, as they will probably change a bit with different materials anyways.... If you want to work in Metric, to get the tensile strength in MPa (which is 1 Newton/sq.mm), set the force gauge to Newtons, and divide by 25 (assuming the test section is 5x5mm, so correct the print to those dimensions).... Attached are the files for both test bars....

Bob

[TorqueMaster]

How high does your force gauge go?  Looking forward to your results!

I did similar, but I was testing the effect of temperature and layer thickness on Z-strength (the weakest direction) for my different filaments, printed at 100% fill.  I used a luggage scale, 50kg max I think, and had to add double pulleys to reduce the force the scale was seeing, even though I used pretty small cross-sections on my specimens.  As I recall I was approaching 300lbs on the specimens, 100lbs on the scale.  PETG layers bond really well!

One factor to consider is -- when printing vertical for Z testing, the area intended to break is very small cross-section, and will print very quickly per layer -- depending on your settings, it may print slower mm/s than usual to maintain "minimum time per layer" settings, or it may print the tiny layers at normal speed, heat soaking the area.  Either way, it may give inaccurate results compared to what strength a normal print, with a decent sized cross-section would have.  The only way I know to avoid that, is to print multiple specimens at the same time, so each one is printed at normal speed, and has a realistic time to cool before the next layer gets added.    Making the cross-section area bigger would also work, but the forces involved would be more than my meager shop tools could handle.

[subscriber]

Excellent plan, Bob Sterne

Excellent strategy, Bob TM

[rsterne]

Yes, I know that the force is likely more than my 500N (50 kg-110 lb) gauge will read, so I am making a setup with (at least) 2 increases in force, using a lever, at 2X and 4X, so I can handle up to 200 kg (440 lbs), or more if I add another hole in the lever.... I'm working on the setup now, and should be able to share that within a few days.... Hopefully, a nice surprise in store for all of you!.... I will be printing multiple samples at a time, so there will be ample time for cooling between layers in the small vertical test section when doing the layers.... I am very curious about how temperature, cooling, and speed affect the layer bonding.... 

Bob

[subscriber]

So, unless I screwed up, a 0.2 x 0.2" specimen that fails at 240 lb has an ultimate strength of 6000 PSI.  If so, that would seem close to the strength of the base material.  Very impressive, if it is applied across the layer plane.

Any strength measure is better than none.  It is probably important to apply the load at about the same rate for comparisons between print direction, or fill percentage, for example.  Very slowly applied force allows the material load to even out via creep.  Very fast, and shock becomes a factor.

Anyway, I have been impressed with printed PETG, both in my hands, and indirectly via parts that Bob has printed, holding up better than I expected.  It has taken me awhile to trust this observation and design lighter parts for 3D printing.   

I think PETG is a more durable material than PLA, despite the latter being stiffer and stronger.  PETG is not as easily degraded by environmental factors; and does not creep so badly when a part is left in a vehicle in the sun.

I am sure that you have seen the video below, or ones like it.  Ultimately you only have to satisfy yourselves that your prints are strong enough for purpose, rather than getting into arguments with forum members or potential customers.

Certainly, it sounds like PETG is a friendlier material to print than ABS; stronger even if the cosmetic results are not as good.  But then my own printing experience is limited to a Kodama Trinus using PLA.  And that was a few years back...

[WhatUPSbox?]

Bob, Stefan from the CNC kitchen videos Subscriber linked has posted a number of 3D print strength comparison tests. He also posted the build of his Instron like DIY tester.

[rsterne]

That's a very cool Tensile Strength Tester.... CNC Kitchen videos are consistently great....

Well, I have been working on the new tester, here is a photo of the progress.... The test bar is in the 2:1 position, with the 4:1 position to its left.... The slotted bar is actually parallel to the base, poor photo.... 



I decided to change the design of the test bars slightly, thinning the "Z" test bar to 9mm instead of 10mm, so that it will fit in a 3/8" milled slot, and lengthening it by 7mm to allow for fillets between the 45 deg. narrowing sections and the 5mm / 0.200" square test strip.... The holes are now 40mm between centers, and the overall length is 62mm.... If a longer test section is needed, just remix the center test portion, it is 5 x 5 mm and by default is 6mm long.... Here are the new test bars....



This should allow a smoother transition of the force into the test section.... The new .stl files are below....

Bob

[rsterne]

I got it finished to the point of operation today....



Yes, it's mounted on my lathe!.... By using 45 RPM and 0.0044" feed per turn, I get 0.198" per minute, which is within a whisker of the 5mm/min. pull speed that is the standard for tensile strength tests (according to the internet).... The main part is mounted in the "T-Slot" on the carriage, by removing the tool post and turning the compound feed parallel to the bed.... I move the carriage so that I can get the test coupon in place, drop in the 3/8" quick pins (cut off bolts), engage the half nuts to the lead screw, and then back the compound feed out until the force meter just moves.... Flip the switch on the lathe, and it pulls at constant speed until the coupon breaks.... I have the force meter set to read peak force, which it holds until you can write it down, and prepare for the new test.... Here is the broken coupon, still in place....



The fixed mount is bolted to a vertical aluminum bar (3/8" x 2") which is bolted to the bed below the lead screw, resting against one of the cast iron cross webs.... I used the 2:1 position for testing a "Z" sample (layer strength, about 2400 psi), and the 4:1 position for the "X" sample, where the filaments are running the length of the coupon.... It broke at just under 6500 psi.... Here is a photo of the coupons....



The top one, the "Z" test failed exactly as it should, between the layers in the narrow test section....

The middle one was the current design of "XY" coupon, and it failed in the middle, but a piece blew off one side....

The bottom one is the original "XY" coupon, and it failed by delamination in the 45 deg. angle portion, rather than in the test portion.... That was the reason for the design change, and it appears I may need to make it even longer, with shallower angles or a larger radius....

All in all, I am VERY pleased with how it works.... 

Bob

[subscriber]

Bob,

You have achieved a sophisticated test system, with minimal investment in new hardware.   Very impressive.

[rsterne]

I wonder what I need for the angle leading into the test section so that it fails first?.... It appears that with 45 deg. it delaminates, which I guess makes sense, as the force to do so could be as low as ~ 2400 psi x 1.414 = 3394 psi.... At 30 deg. then it would be ~ 2400 x 2 = 4800 psi.... and at 20 deg. it should be ~ 2400 x 2.92 = 7000 psi, which is more than the tensile strength of ~ 6500.... Those numbers are, of course 1 / sin(angle), which I think is the force pulling the layers apart at that angle, isn't it?....

I'm going to make some XY coupons with different angles until the failure is always in the test section, hopefuly I'll learn something along the way....

Bob

[subscriber]

That angular consideration is an interesting question, Bob.  At what point does someone argue that your coupons are not standard?  Unless you are talking about leaning the standard coupon relative to the printer axis by so many degrees.  If the latter, leaning the coupon surely means that the test section is also not orthogonal?

My attitude about a standard coupon or test method would be, they are a starting point, until you develop something more sensible.  Then, maybe others might adopt a new standard for testing 3D printed parts.

[WhatUPSbox?]

Bob, great work.
I found some light reading for when your next set of coupons are printing. Some of the discussion deals with optimizing the print path to avoid some of the failure modes you've seen. Of course, that gets into the question of staying representative of a typical (unoptimized) 3D print. In general, the ASTM dog bone coupon configurations tend to be longer than your initial coupons.
https://www.osti.gov/servlets/purl/1669703 (pdf)
https://www.mdpi.com/2073-4360/15/14/3029
https://www.sciencedirect.com/science/article/pii/S2238785423010578

Looks like a fun project.

[rsterne]

The ASTM coupons require a clamping system, do they not?.... I want to use the through hole type (quick and easy replacement with quick pins), so I will have to work on the center section and lead-ins to it until it breaks reliably only in the test section.... Four versions are on the printer as I type!....

It is easy to make a longer section for the XY coupons, but not so much in Z.... For example, I don't think you could print a standard ASTM Dog-Bone in Z.... I think the Z shape I have seems to work fine (pending more tests), as in reality all it it doing is testing the interlayer bonding, so the failure should always occur at the smallest area (unless there is a flaw in the print)....

Bob

[WhatUPSbox?]

I think your configuration would accommodate a clamping approach outboard of the pins. The clamp approach would open up the geometry options and keep the stress field closer to the ASTM standard. I am thinking of a functional equivalent to the simple clamps used in the test stand from the company that makes your force gage.

www.amazon.com/Mxmoonfree-ZMF-500N-Force-Gauge-Manual/dp/B0C7VBB84D

[subscriber]

I like a single pin at both ends of a tensile sample:  The pins make effective hinges so the system is self aligning.  This avoids bending of the sample.  Else, you have to grip the sample very carefully to avoid misalignment that could induce bending.  The rest of the system also has to have good alignment, and constraints to avoid moving off axis.  Constraining the axis could induce friction, unless you have ball slides. 
Hinges are so much simper...

That amazon stand is amazingly cheap, but it has you crank by hand to produce the test force.  Not a bad option, but the speed at which the force is applied is also part of the test standard.  If you can't exactly match that standard, being consistent is a pretty good second.  Bob's lathe setup is not desktop friendly, but it is consistent.

All that said, modifying the amazon test stand to add hinges to the jaw arrangement would take away my objections.  Ball joints would be better, as they hinge in all directions.  Or two hinges at 90 degrees, like a universal joint.

[rsterne]

Yes, I could no doubt change both the test equipment and the coupons, to be closer to the ASTM standard, but in truth I don't care.... My object is to get repeatable results to be able to explore how changes in printer settings change the results, rather than what the actual tensile strength (as per ASTM) is.... I of course want to be able to compare different filaments as well.... Anyways, today I learned more than I could have imagined, just with a few tests, all done with PLA....

First I tested the effect that changing the number of walls had.... Before I built the machine, I had already made some coupons, from the original Thingiverse design, at 5mm thick, with a 45 deg. angle.... I used 2, 4 and 8 walls, expecting that when the test section travel was completely aligned with the stress (ie all walls) the strength would maximize.... Well, that didn't occur.... Two walls at the top, 4 in the middle, and 8 at the bottom....



Here are the numbers, time to break (in seconds), stress (distance stretched at break in mm), strain at break (maximum in lbs.), and tensile strength in psi.... plus mode of failure....

2 walls - 67 sec.... 5.58 mm.... 327 lbs.... 8170 psi .... Broke across center
2 walls - 66.... 5.50.... 323.... 8070.... Broke across center
4 walls - 73.... 6.08.... 342.... 8550.... Broke across center
4 walls - 75.... 6.25.... 350.... 8760.... Broke across center
8 walls - 64.... 5.33.... 271.... 6780.... Delaminated along 45 deg. shoulders
8 walls - 56.... 4.67.... 283.... 7080.... Delaminated along 45 deg. shoulders

My conclusion is that the infill, which is 100% density of diagonal lines alternating on the 45's, is preventing the two sides of the test section from moving relative to each other, causing the delamination along the sloped shoulder.... Next, I tested a series with different shoulder angles, but all with 8 walls, expecting to see less delamination with the shallower slope, which did in fact occur.... Description below is from top to bottom in the photo....



At 30 deg. the strength was significantly more than at 45 deg. (top coupon), but it still failed when the slope delaminated.... With 25 deg. slopes (and 20 deg.) the tensile strength increased to be in between the 2 and 4 wall tests in the previous section, and the coupons broke across the middle.... The bottom two coupons are a 30 deg. shoulder angle with a longer test section, and a new coupon design with round ends, a 30 deg. shoulder, and a radius between the shoulder and test section.... They performed almost identically, and failed slightly before the 30 deg. sample 2nd from top in that photo.... Interestingly, all six coupons showed some delamination, whether that was the method of failure or not.... Also, all coupons tended to delaminate along the centerline....  .... Here are the numbers....

45 deg.... 43.... 3.58.... 220.... 5510.... Delaminated
30 deg.... 57.... 4.75.... 300.... 7500.... Delaminated
25 deg.... 71.... 5.92.... 331.... 8270.... Broke across center, some delamination noted
20 deg.... 66.... 5.50.... 333.... 8330.... Broke across center, some delamination noted
30 Long.. 57.... 4.75.... 282.... 7040.... Delaminated
30 Rnd... 57.... 4.75.... 282.... 7060.... Delaminated

My conclusion is that for this style coupon, with these printer settings, a slope of 25 deg. or shallower is needed on the shoulder, and that there is no significant difference in the strength or failure method using a rounded coupon or radiused filled into the test section (but it uses less filament and printing time).... There appears to be a lack of bonding between the inner walls of the test section, allowing the two sides to split apart.... Possibly this is due to using 0.4mm walls (a total of 12) which take up 4.8mm in a 5mm wide part.... I am going to try using a slightly thicker wall to get the inner walls to bond better, or possibly increase the percent extrusion slightly about 100%.... I am not sure, bit I think that should tend to produce "square" filament strands instead of round one?....

By the way, my method of timing is to set the coupon up with minimal clearance to when the lead screw starts to pull on by using the compound feed, so that the starting load on the gauge is zero.... I then turn on the lathe, and start a stopwatch when the numbers start to increase, stopping it when the part breaks.... Interestingly, in the first few seconds the load increases slowly with slight pauses, until the coupon settles into full engagement with the quick pins.... The load then rises in a pretty linear fashion until a few seconds before failure, when the stress plateaus and the coupon is yielding before it breaks....

Bob

[Madd Hatter]

When I did tensile testing I use to load at .050 inches per minute. Any idea of your load rate? Of course my ultimate load use to be around what your psi results. Is your pin holes slightly rounded so as you are loading the coupon you're not putting any twisting load into it?

[rsterne]

Robert, I am loading at 5mm (0.2") per minute, which I read somewhere is the standard load rate?.... The holes in the coupon are straight sided, but all the holes in the test machine were done on a mill, to they are as parallel as humanly possible.... There is vertical clearance on the pins so that the coupon can slide up and down to align itself horizontally with the load.... The results seem to be very consistent, so I think accuracy of the machine and coupons is not problematic....

Bob

[Madd Hatter]

Most of the tensiles I tested was at.050 but there were some at .2 but it's been more than a decade since I last did any testing and can't remember what material was tested at .2.

[rsterne]

OK, so I tried changing the flow rate because I thought I was getting incomplete bonding between the wall layers.... I tried 110% and 120%, but since I was printing inside to outside, the coupons got slightly wider, with smaller holes, which I had to drill out to fit over the quick pins.... The 110% did increase the tensile strength of the coupon slightly (yes, I used the larger areas), but it dropped again at 120%, and the top surface was so rough I thought it might damage the nozzle, which was rattling over it!.... Here are the photos....



The top row is 100%, middle 110%, and bottom 120% flow.... I tried one more pair of coupons with only the inner walls set to 110% flow, all the other settings the normal 100%, and they gave the best results.... No photo, sorry, but here are the best tensile results to date....

30 deg. tapered.... 55 sec.... 4.58mm.... 304 lbs.... 7501 psi.... Small delamination on one shoulder....
30 deg. rounded... 68 sec.... 5.67mm.... 356 lbs.... 8785 psi.... Broke at end of test section....

These looked essentially identical to the middle row in the photo above.... I did see one thing when examining the breaks.... The bonding WAS better with the increased flow rate, which resulted in the highest strength to date....

Bob